OrangeBot.AI Digest — 2026-10-16

Ultrafast workspace for agents to run in parallel

Edit your live UI in your design system, hand off the diff

The hypercar that flies

Open Source Paper / Figma Alternative

E-signatures without the subscription. Pay per doc.

Your entire video library: organized, searchable, understood

Every coding agent, every machine, one command center.

The Dynamic Island to-do app you don't have to open

Open-source desktop app and VS Code panel for Muse Code

Turn any website into an API; Faster & reliable browser use

Free AI NVR that finds any person or vehicle in seconds

A self hosted home page for everyone in the house

SaaS where a user can can book appointments to other users.

The studio your AI coding agents work in

Turn what your team learns into everyone’s agent setup

Policy-shift distillation reuses the difference between a teacher's log probabilities before and after reinforcement learning (RL) as supervision on student-generated trajectories. However, this endpoint policy shift cannot reveal whether local action preferences evolved consistently during teacher training. The same endpoint shift can result from early and late shifts that reinforce or partially cancel each other. We investigate whether agreement between early and late teacher policy shifts can guide selective transfer. We introduce Waypoint-guided On-Policy Distillation (Waypoint-OPD), which uses an intermediate teacher checkpoint from the same RL run to compare early and late shifts in relative action preferences at student-visited prefixes. We gate the endpoint shift according to the directional agreement between the early and late shifts, suppressing it when they point in opposing directions. Compared with endpoint-only transfer, our method requires only one additional frozen-teacher scoring pass. Using RL-trained Qwen3-4B teachers, we evaluate our method with Qwen3-8B, Qwen3-14B, and Qwen3-30B-A3B as student models in mathematical and code reasoning. In all six domain--student settings, Waypoint-OPD achieves the highest domain-average score among the compared distillation methods. Ablations with the Qwen3-8B student show that removing or shuffling gates reduces mathematical reasoning performance. These results support the use of directional agreement between early and late teacher policy shifts to weight the endpoint shift during student training.

Computer use has reached a turning point: the focus is expanding from GUI-only interaction to systems that complete tasks on users' computers through visual and programmatic interfaces. We define a generalist computer-use agent (GCUA) as a system in which the model chooses how to observe, act, and organize its work across visual and programmatic interfaces. Yet frontier implementations, such as GPT-6 Astra's computer-use system, remain largely closed, limiting systematic study of their design. We introduce Open Generalist CUA (OpenGCUA), an ongoing project to develop and open the full stack for GCUAs with the community. This report presents the first release of OpenGCUA. Code execution is a key feature of our agent design: it lets intermediate results determine subsequent operations and enables models to apply their existing programming capabilities directly to computer interaction. We also propose harness-to-model decision transfer, moving selected observations and task-state decisions from fixed harness rules to the model. On OSWorld-v2, our full configuration improves task scores across all eight evaluated models, including GPT-6 Astra and Claude Opus 5, raising mean partial score by +19.6 points and reducing reported steps per task by an average of 53.0% relative to batched GUI-only execution. We release our implementation and prompts, evaluation trajectories from all eight models, and preliminary versions of a real-device harness and a cross-device SDK. The harness addresses real-device challenges not fully captured by benchmarks, while the SDK provides a common programming interface across computers, mobiles and browsers. Across four real macOS tasks under our real-device harness, selected Qwen3.8-Max runs use 19.4 times as many reported reasoning tokens as GPT-6 Astra, highlighting substantial differences between models in real-device execution efficiency.

Existing agent benchmarks typically evaluate agents on well-specified tasks defined by issue reports, specifications, or task tickets. Real-world customer delivery is different: forward-deployed engineers (FDEs) must clarify ambiguous requirements, work in live systems, and uncover unstated constraints. We introduce FDE-Gym, a benchmark of 95 end-to-end delivery tasks from 20 customer scenarios across 18 sectors, each adapted by industry experts from real-world delivery projects. In simulated customer environments with scripted stakeholders and modifiable customer systems, agents interact with stakeholders and operate on production-like systems to deliver working solutions. FDE-Gym scores a delivery by how much it improves the customer's own success metric, normalized between 0 (unchanged system) and 1 (reference solution), with penalties for stakeholder interactions (messages, trial runs, and waiting time). An online platform lets humans practice FDE skills under the same grading criteria. We evaluate 13 model configurations from seven providers. In the realistic setting, where unstated constraints must be discovered, the best configuration achieves a mean FDE-Gym score of 0.56 and resolves only 14 of the 95 tasks (a score of at least 0.8 with no incident). Discovering unstated constraints is the primary bottleneck: performance drops sharply when models must uncover them without explicit guidance. Yet information discovery alone does not account for the gap: even on the 23 tasks where the best configuration obtains every key fact from stakeholders, it reaches a normalized business uplift of 0.8 or above on only 10. These results reveal a substantial gap between solving well-specified tasks and delivering effective AI systems for real customers.

Can vision-language models turn 2D observations into 3D understanding that supports precise action? This requires inferring object geometry, anticipating the effects of rotation, and refining actions through visual feedback. We introduce RoTA-Bench to evaluate these abilities through interactive 3D-to-2D projection alignment. Given a target silhouette, a model observes a rotated object and chooses successive rotations to recover the target projection within a limited action budget. Our automated pipeline constructs silhouette-preserving 3D objects from 2D icons and uses reversible initialization to guarantee solvability, yielding 674 tasks and a stratified 100-task test set. Empirically, we find that (i) a substantial human-model gap persists: GPT-5.6-Sol succeeds on only 17% of tasks, whereas 20 undergraduate participants collectively complete the test set four times with a 100% success rate; (ii) image-space and pose-space progress can diverge: at matched states, the model more often improves silhouette overlap while worsening pose alignment than humans (19.4% vs. 10.9%); and (iii) fine-grained control is sensitive to interaction history: removing it leaves 96.8% of the model's actions at a coarse 30-degree step size and sharply reduces observed task success. These findings highlight the challenge of converting visual evidence into precise 3D control, beyond recognizing or matching individual images.

Harness self-evolution revises an agent's software scaffolding, including prompts, memory, workflows, and executable components, while keeping model weights fixed. We formalize a feedback-driven process in which accepted revisions persist across runs. The framework separates update targets from proposal and admission mechanisms. It distinguishes changes to interpreted state from revisions to tools and code, repository-level control logic, and plugin composition. We characterize search through evolutionary optimization and examine its reliability through continual PAC-learning. Under explicit assumptions, the analysis yields an expected generalization bound, a retention bound conditional on historical regression budgets, and a high-probability admission rule using fresh, independent validation data. We examine how persistent harness revisions affect knowledge sharing and organizational workflows, with implications for scientific reliability, physical safety, and institutional accountability. Cases from prior work illustrate model--harness co-design, distributed execution, and plugin reuse. Reliable deployment requires continual validation as task distributions and evaluators change, with independent safeguards for learning, runtime privileges, and consequential actions. We provide a comprehensive 66-page discussion of related work, theory, algorithms, evaluation, applications, and the broader impact of Harness RSI, together with continuously updated code and supplementary materials, at https://github.com/zgcagi/Harness-RSI.

Spatial intelligence lets an agent perceive and act in the physical world, and many of its judgments must be made at interaction rates: which spatial relation holds, how far an object is, or where to place it. Vision-language models (VLMs) produce these answers by free-form text generation, paying for autoregressive decoding and answer parsing even though the output space is compact: one of a few options, a scalar, or a 2D point. In light of this, we introduce MJev-Spatial, a System 1 multimodal decision model that tackles all three types of spatial understanding tasks within a unified classification framework. Specifically, MJev-Spatial treats each candidate as a class and scores it using the backbone's pretrained query and key projections, without adding parameters. The candidates are the given options for choice tasks, or value bins and image regions refined through coarse-to-fine classification for numerical and point prediction, respectively. At inference, each classification decision takes a single forward pass and generates no answer tokens. To train on heterogeneous spatial data efficiently, we supervise all coarse-to-fine stages jointly with teacher forcing in one forward pass, and combine this with length-aware global batching and cross-GPU load balancing, raising training throughput by 30.8% over the original batching strategy. Built on Qwen3.5-4B and trained on 12.3M question-answer pairs converted from existing spatial datasets, MJev-Spatial achieves the highest overall score (65.64%) among eight tested models, ranks first on seven of twelve benchmarks, and outperforms its unmodified autoregressive base model on all twelve, while running 2.5 times faster than the fastest autoregressive baseline and 5.6 times faster than the base model.

Online reinforcement learning (RL) with verifiable rewards has driven rapid progress in mathematics, coding, and tool use, but scaling this success to GUI agents remains challenging due to the high cost of interactive environments and the scarcity of realistic training tasks. We introduce MAI-Gym, an open all-in-one platform that unifies GUI agent evaluation, data collection, and online RL training across mobile, computer, and web environments. Through a shared rollout pipeline, MAI-Gym ensures consistent agent behavior across evaluation and training, supports ten established benchmarks, and scales to 20,000 parallel environment instances. Beyond scalable infrastructure, MAI-Gym enables automated synthesis of diverse GUI applications and challenging tasks, including resettable mocked apps and coherent cross-app worlds with execution-validated task verifiers. Using MAI-Gym, we train MAI-Gym-9B, a unified 9B agent for mobile and computer use, achieving state-of-the-art performance among open GUI agents under 10B parameters: 78.4% on MobileWorld and 76.2% on OSWorld-Verified. Importantly, RL in a mocked train-ticketing app transfers to its real-world counterpart, improving task success on a phys- ical phone from 28.8% to 48.1%. We release the framework, synthesis skills, mocked apps, 2,000 verified tasks, and the model checkpoint to support repro- ducible research and further advances in scalable GUI agent training.

Vision-language-action (VLA) models transfer semantic knowledge acquired through vision-language pretraining to robot control, achieving strong performance across various benchmarks. However, most VLAs directly map observations to demonstrated actions without explicit supervision of intermediate physical state changes, making them prone to overfitting observation-action correspondences. Some methods introduce intermediate reasoning or future-state prediction in language, visual, or latent spaces to guide action generation, but do not explicitly supervise the physical state transition itself. We therefore propose VLA-Δ, a general framework that uses physical changes between current and future states as explicit learning targets. To support this objective, we construct a large-scale pretraining dataset and derive physical state-transition annotations covering task-relevant semantics and visual motion changes, providing more direct supervision for continuous action generation. During inference, the model generates actions directly from current observations without additional inference overhead. VLA-Δ achieves performance competitive with the state of the art on LIBERO and RoboTwin, and outperforms π_{0.5} in real-robot experiments. Our code is available at https://github.com/ZJU-OmniAI/VLA-Delta

Most existing agentic reinforcement learning methods assume single-session trajectories, where interaction proceeds within one continuously accumulated context. However, agentic techniques such as context compaction and subagent forking may produce trajectories spanning multiple sessions, creating a mismatch between the single-session training assumption and inference-time execution. Therefore, we introduce Multi-Session Agentic Reinforcement Learning, a more general setting that supports an arbitrary number of sessions within a trajectory. Under this setting, a widely adopted method is to broadcast trajectory-level advantages to all sessions, but this does not distinguish contributions from various sessions or guide session allocation within trajectories. To address this issue, we propose Mosaic, a multi-session agentic RL method with critic-free hierarchical optimization. At the trajectory level, it incorporates session-usage signals into advantage estimation. At the session level, it shapes each session's advantage according to the estimated marginal contribution. Experiments on BrowseComp with Qwen3.5-4B show that Mosaic outperforms broadcasting in both context compaction and subagent forking scenarios while using fewer sessions on average. Further analyses show that our method achieves improvements in session utility and allocation, which we suggested is a promising target for future advances in multi-session agentic RL.

World and action modeling are closely related, yet their reciprocal benefits under unified causal training remain unclear. We introduce OneWM, a simple framework for unified causal training that treats actions as both conditions for world prediction and targets for policy learning within a shared video-action architecture. Using this framework, we investigate whether learning forward dynamics conditioned on specified actions improves action learning, and whether learning to generate actions in turn improves action-conditioned world modeling. We use LIBERO, LIBERO-Plus, and RoboCasa365 as embodied testbeds to evaluate both policy learning and world modeling, complemented by real-world policy experiments. Learning the visual consequences of specified actions improves policy performance and generalization across changing scenes. In turn, action learning improves multi-view world-prediction fidelity under the same architecture. Together, these results answer both questions positively, supporting reciprocal benefits between world modeling and action generation under unified causal training.

Humans interpret multimodal observations of the world and forecast how it will evolve. Recent multimodal generative systems have demonstrated similar capabilities: they can integrate multimodal context and synthesize visual outputs that depict changes in the world. To characterize this emerging trajectory, we organize model capabilities into a four-level hierarchy and discover that higher-level tasks remain relatively underexplored despite model progress. To bridge this gap, we explore model performance on multi-step world state prediction, a high-level task that requires models to forecast successive future world states from multimodal observations and diverse references. Such a task requires models to possess entangled understanding and generation capabilities, as well as extensive world knowledge to translate human action sequences, physical transformations, and abstract visual reasoning into coherent visual outputs. We introduce WorldCast, a benchmark comprising 320 manually designed cases across eight categories, and evaluate 12 representative models. The conducted experiments yield three key insights: (a) advances in frontier models are rapidly pushing the boundaries of intelligent world modeling, with the GPT-Image-2.5 System scoring 72.8 and demonstrating early-stage visual state prediction capabilities on real-world tasks; (b) WorldCast reveals substantial gaps between models: even Nano Banana Pro reaches only 42.8 and the best open-source model scores 20.1, with deficits in handling complex context, entangled references, and reasoning-heavy tasks; (c) strong generative models effectively draw on the capabilities of their underlying language models. These findings highlight the need for multimodal generative models that are capable of dynamic planning and reasoning while also being deeply unified to faithfully simulate real-world states.

Generative visual verifiers play a crucial role in advancing multimodal intelligence. To enhance the robustness of visual verifiers in complex multimodal scenarios involving compositional requirements, we move beyond conventional binary verifiers that produce only True/False discrete judgments and introduce a verifier that automatically decomposes complex prompts or environmental rules into fine-grained components and assigns pointwise scores to provide a continuous and more informative critique signal. Our experiments further demonstrate that these structured and interpretable outputs improve the verifier's generalization to out-of-domain multimodal critic tasks. To train this verifier, we introduce a Decoupled RL training strategy that jointly optimizes sub-question-level binary judgment and global alignment scoring, using precise local feedback to develop fine-grained verification capabilities that provide a foundation for reliable continuous scoring at the global level. Under a general reinforcement learning framework, we first empirically demonstrate that the proposed pointwise graded verifier significantly outperforms binary verifiers in traditional text-to-image scenarios. We then extend our investigation to rule-governed generative world modeling where generative models are leveraged for world simulation and world reasoning under explicit rules. From a theoretical standpoint, we further show that a pairwise verification paradigm yields more accurate critiques than the pointwise formulation when ground-truth images are available, with empirical results on diverse structured world-modeling tasks such as Maze and Sudoku further validating the effectiveness of our approach. Overall, our findings suggest a key insight: the transition from generative models to structured world-modeling agents critically hinges on the availability of accurate pairwise graded visual verifiers.

Data agents use code to process heterogeneous files and derive analytical insights. Scaling their training faces costly environment curation, privacy and intellectual-property restrictions, and difficulty verifying open-ended findings without known generative truth. Our core insight is to specify the world behind the files before generating the files themselves. We introduce Data World, a multi-agent engine that synthesizes environments, tasks, and rubrics from explicit mechanisms. Independent audits check target recoverability from observed files; rubrics support evaluation and RL rewards. Human preference-based curation for question realism selects 100 tasks across 50 held-out worlds and 16 domains for DataWorld-Bench. The engine also yields 2,135 training tasks from 1,037 disjoint worlds. Evaluation of ten frontier models shows that the benchmark remains challenging, particularly for insight discovery, with a consistent computation-insight gap. RL-zero training on the synthesized tasks produces DataWorld-35B-A3B from Qwen3.6-35B-A3B without trajectory distillation. Under matched evaluation, it improves over its base by 7.06 points on DataWorld-Bench, 14.28 on DAComp, 20.90 on DABstep-Hard, and 12.13 on JobBench, a broader occupational benchmark.

Large language models (LLMs) are expanding from assisting scientific research and engineering to autonomously developing and improving AI systems, advancing AI for AI (AI4AI) as a pathway toward recursive self-improvement (RSI). Workflow-driven machine learning engineering (MLE) systems provide a practical path toward this pursuit through complex orchestration that manages long-horizon experimentation, while recent MLE systems adopt model-directed interaction, allowing models to autonomously organize experiments and adapt optimization strategies based on execution feedback. However, feedback from each experiment reflects several preceding decisions, and a long MLE run contains many such costly experiments. Training must therefore select useful supervision from long interaction trajectories and learn from intermediate outcomes. We propose a post-training method for autonomous long-horizon MLE optimization and apply it to train MLExpert, a 35B Mixture-of-Experts agentic model. We design the Kaggle-MLE Factory to construct executable tasks and an interaction harness for model-directed experimentation. Hierarchical agentic SFT transfers general tool-use and coding capabilities to MLE through outcome-guided supervision. Windowed Interaction RL (WIRL) treats the multi-turn decisions between successive experiment results as an execution window and assigns it a shared outcome-based reward. Asynchronous collection and policy correction support policy updates before the trajectory ends. On MLE-Bench Lite, MLExpert improves the average medal rate of the base model from 37.9% to 72.7%, achieving performance comparable to larger frontier models such as Kimi-K3 and GLM-5.3. Further experiments show that MLExpert improves performance on NatureBench Lite, agent scaffold optimization, and algorithm optimization, suggesting that the learned optimization capability generalizes to other tasks.

Language models are increasingly trained against several verifiable rewards at once, such as the correctness and the format of a tool call, or the individual constraints of a complex instruction. Reward-decoupled normalization (GDPO) normalizes each reward separately so that different reward combinations keep distinct advantages, but it still spreads every reward's advantage evenly over the whole response, although each verifier checks only a few positions. We propose MRTPO, which assigns the advantage of each reward to the tokens relevant to it, using the information its verifier checks, without auxiliary models or token-level annotations. For each reward, the policy replays its own response with and without that reward's verifier information, and the per-token Jensen-Shannon distance between the two next-token distributions measures each token's relevance and reassigns the advantage. On text-to-SQL, multi-constraint instruction following and tool calling, MRTPO converges faster and achieves better performance than the compared methods.

Mark Gurman / Bloomberg : Sources: Apple has hired Scott Tucker, Meta VP and deputy general counsel, as its new litigation chief, as part of a shakeup in its legal team —  Apple Inc. has tapped a longtime Meta Platforms Inc. executive as its new litigation chief, bolstering its legal team at a time when the iPhone maker faces a growing array of lawsuits.

Michael Acton / Financial Times : Apple says demand for iPhone Duo has been “nothing short of extraordinary”; deliveries slipped to late January within hours of preorders opening —  Deliveries of the $2,000 Duo slip to 2027 within hours of going on sale  —  Apple on Friday said its foldable iPhone Duo had seen …

Theodore Schleifer / New York Times : Filings: Meta reduced its $65M midterms budget by $25M through refunds, primarily from a super PAC it launched in 2025 with $45M to fight state AI bills —  The company outlined a $65 million budget last year when it embarked on an effort to elect state politicians friendly to the A.I. industry.

Anthropic : Anthropic says Claude traced NIST's ECDSA curve seeds to a pseudorandom number generator, challenging decades-old suspicions of an NSA backdoor in the algorithm —  “It's just timestamps”: the decades-old mystery of the origins of one of the most popular cryptographic standards has been explained.

Bloomberg : Sources: SoftBank weighs buying US security services giant Allied Universal at an ~$18B equity valuation, as Masayoshi Son explores deploying AI-powered robots —  SoftBank Group Corp. is considering an acquisition of Allied Universal, the largest security services company in the US …

Joe Barnes / Telegraph : Sources: Elon Musk approached Trump with the idea of slowing down Starshield in Ukraine; source: Trump agreed, paving the way for DOD to execute the slowdown —  SpaceX founder had publicly blamed president for chaos over restrictions, but sources say he initiated them

Lindsay Clark / The Register : The PostgreSQL Global Development Group delays the general availability of PostgreSQL 19 after LLM-assisted code reviews uncovered architectural defects —  Core devs revert up to 10 features, push release to November, and blame better tooling rather than vibe coding

Dylan Sloan / Bloomberg : Three CoreWeave co-founders have sold $3.3B in stock since its March 2025 IPO, including $2.4B in 2026 alone; CoreWeave stock fell 60%+ from its June 2025 peak —  Three CoreWeave Inc. co-founders have sold $3.3 billion in company shares since last year's initial public offering …

Matt Marshall / VentureBeat : Reflection AI releases Apache 2.0-licensed weights for Beam, a 501B-parameter model, trained with RL that penalized unnecessary tokens —  Reflection AI today released the weights of Beam, its first open model, and is pitching it against the Chinese open models that many U.S. companies …

John Harney / Bloomberg : In its first meeting, Trump's Super Intelligence Force says it plans to commission a report on AI risks and develop protocols for labs to provide early warnings —  President Donald Trump's artificial intelligence task force in its first meeting agreed on Thursday to address risks related to cybersecurity …

Jess Jones / The Times : Source: Intuit acquired UK-based B2B revenue management and billing startup Sequence in a sale worth ~$200M; Sequence raised $19M led by a16z in 2022 —  A British financial technology business has been acquired by Intuit, the American software company, in a sale estimated to be worth around $200m …

Andrew E. Freedman / Tom's Hardware : Microsoft Surface Laptop Ultra review: powerful, useful Magnetic Connect port, bright screen, but heavy, loud, and Windows doesn't take advantage of RTX Spark —  Useful for AI models, but still waiting for more Windows 11 features to take advantage of.  —  Reviews

Aaron Tilley / The Information : Sources: PayPal, Epic, and others voiced concerns to DOJ as it signaled a preference to settle its Apple antitrust suit with a deal focusing on interoperability —  As the U.S. Department of Justice negotiates a settlement of its antitrust lawsuit against Apple, app developers are fretting …

Bloomberg : Organizers abruptly cancel 2026's Grace Hopper Celebration, one of the biggest events for women in tech, after sponsorships and registrations fall short —  The cancellation of this year's Grace Hopper Celebration, one of the largest gatherings of women in technology, is the latest example …

Prathik Jayaprakash / Reuters : Waymo plans to launch its robotaxis in Toronto in 2028 in partnership with Element Fleet Management, the company's first deployment in Canada —  Google's (GOOGL.O) Waymo said on Friday it plans to bring its autonomous ride-hailing service to Toronto in 2028, marking the self-driving company's …

英国中年男性自杀率达历史最高水平。英国国家统计局 2025 年的数据显示，50 岁出头的男性自杀风险最高。2025 年登记的自杀死亡人数为 6162 人，略低于 2024 年的 6190 人，总体自杀率基本不变，即每 10 万人有 11.3 人死于自杀。接近三分之一死者为 45-64 岁男性，该年龄段自杀人数比 2024 年增加 49 人，该年龄段自杀率达到每 10 万人 23.8 人，高于 2024 年的 23.2 人，为 1981 年开始统计该数据以来的最高纪录。其中 50-54 岁男性自杀率达到每 10 万人 28 人。有 553 名中年女性死于自杀，每 10 万人 7 例，于 2024 年持平。Rethink Mental Illness CEO Mark Winstanley 称自杀原因很复杂，许多人正承受着巨大的生活压力，包括经济困难、就业不稳定、住房成本上涨、孤独感以及心理健康问题。

Meta 发布了 Muse Spark Open，称它是目前美国最强大的开放权重模型。Muse Spark Open 为多模态模型，总共有 8160 亿参数，激活参数 420 亿，支持 100 万 Token 上下文。该模型利用了苹果最新版本的开源数组框架 MLX，对苹果 Apple Silicon 的统一内存架构进行了优化，能运行在搭载 M5 Ultra 芯片的 Mac Studio 上。M5 Ultra 支持最大 512GB 内存容量，两台最高配置的 Mac Studio 就能容纳整个模型，两台机器相当一台 1TB 统一内存的计算机。Muse Spark Open 已发布在 Hugging Face 上。

日本京都大学的研究人员在 iscience 期刊上发表了一篇论文，他们在一个月内测量了 19 名主人和 29 只宠物猫的催产素和皮质醇水平，评估主人的情绪，通过蓝牙信号监视主人和猫之间的距离。结果显示，主人与猫的催产素水平呈正相关，皮质醇水平之间的关联很小。猫的催产素水平越高，与主人皮质醇水平就越低和负面情绪越少相关，主人更高的皮质醇水平与更多的负面情绪相关。主人与猫之间距离越近，主人的活力水平就越高。探索性行为测试则将猫的催产素与社会接近行为联系起来，将皮质醇与对陌生人的接近减少联系在一起。与以皮质醇同步为中心的人犬二元关系不同，人猫关系可能以催产素为核心，将跨物种的激素研究扩展到一种具有独特驯化历史的物种。

2000 年生物学家史蒂文·奥斯塔德提出了一个颇具争议的说法：“世界上第一个将活到 150 岁的人现在很可能已经出生了。”他当时乐观地认为，科学将找到延长寿命的新方法，而且他至今仍持这一看法。似乎数不胜数的网红和未来学家也抱有同样的信心，其中一些人甚至做出了更为惊人的预测。生物医学老年学家、SENS 研究基金会创始人奥布里·德格雷声称，如今 40 岁的人有超过 50% 的可能性永远不会因衰老而死亡。但今天研究人员对于未来究竟能在多大程度上延长寿命仍争论不休。问题在于，他们的答案往往只是假设，而且无法得到验证。归根结底，单靠某一项突破性成果恐怕无法阻止衰老。比如大多数肌肉细胞和脑细胞都无法自我更新。就目前而言，我们无法活那么久的最有力证据，可能就是一个简单的事实：有记录以来，还没有人活到 150 岁。自 1997 年以来，有记录可查的最长寿命纪录一直没有被打破。那一年，法国女子让娜·卡尔芒去世，据称享年 122 岁。她的真实年龄存在一些争议。“自那以后，甚至没有人活到 120 岁，”奥斯塔德博士说。

微软表示，Windows 11 最近的更新通过改进内存管理和减少后台资源占用，系统响应速度提升最高 20%。改进内存管理也显著改进了只有 8GB 内存的电脑的运行表现。由于内存价格飙升，越来越多的 PC 厂商重新开始推出只有 8GB 内存的电脑。微软 Windows 体验部门企业副总裁 Navjot Virk 表示：“Windows 现在可在内存中多保留多达 2GB 的活动数据，而不是把其转移到更慢的存储设备。比如在内存为 8GB 或更少的电脑上，组件（Widgets）只在你打开时才会加载，这可以为应用和其它工作释放出逾 200MB 的内存。”Windows 11 的下一个 重要更新是新的 Windows 搜索体验，速度更快、功能更强。目前已面向 Windows Insider 用户开放，微软将在今年晚些时候推送给更多用户。

根据 IDC 的数据，三季度全球智能手机出货量同比下降 22% 至 2.553 亿部。受内存价格上涨影响，手机价格上升，出货量降幅创历史最高。IDC 预测未来智能手机市场还将进一步下滑。根据 IDC 的数据，三季度中国智能手机市场出货量约为 5790 万台，同比下降 15.4%。华为是唯一实现同比双位数增长的头部品牌。华为份额升至 21.1%，同比增长 17.3%；其次是苹果的 17.8%，同比下跌 4.3%；之后是 OPPO、vivo、小米、荣耀、WIKO、联想（市场份额 0.4%，同比增长 1.1%）、三星和中兴。除了华为和联想，主要手机厂商的出货量都下跌。

日本宇宙航空研究开发机构（JAXA）将于 10 月 20 日凌晨发射火星卫星探测器“MMX”。任务目标是登陆火星卫星“火卫一”并采集样本。MMX 将历时约 1 年抵达火星圈，并登陆火卫一。任务计划利用机械臂采集 10 克以上土壤。预计于 2031 年返回地球。法国和德国联合研发的探测车将先行登陆火卫一。该探测车搭载于 MMX 上，将与其分离并投放。探测车为边长约 50 厘米的立方体，重约 23 公斤。探测车将通过车轮在火卫一表面移动，调查温度、成分等，为 MMX 采集样本提供支持。此次探测的最大目标，是解开火卫一如何形成的谜团。目前较受重视的理论包括：来自其他区域的小行星被火星引力捕获；以及巨型陨石撞击火星时飞散的岩石聚集形成火卫一。调查火卫一砂样，很可能为这一争论画上句号。若其源自小行星，或留存有太阳系诞生初期的物质，有望帮助揭示行星形成及生命诞生的过程。

通过研究对 50 种鸟和蝙蝠的跨越 17 年的观测数据，研究人员提出的见解挑战了传统观点——即飞行具有可预测的能量成本，且总是高于游泳或休息等其他移动方式的成本。这项新研究表明，鸟类和蝙蝠可以利用风力和上升气流来大幅减少飞行所需的能量。研究人员测量了真实条件下的有效运输成本——即飞行动物移动给定距离所消耗的能量，分析了 50 种鸟和蝙蝠的数据，这些物种的体型跨越了三个数量级（从约 12 克到超过 11 公斤）；研究使用了来自 850 只动物的历时 17 年的高分辨率 GPS 和加速度传感器测量数据，并辅以针对小型物种的雷达观测数据。研究结果表明，动物在野外飞行时的能量成本在很大程度上取决于它们如何与大气相互作用。翱翔的鸟类（尤其是体型较大的物种）可利用上升气流和其他大气气流来大幅降低能量消耗。在某些情况下，这种方式能使飞行成本降至与游泳甚至静息状态相当的水平。相比之下，频繁扇动翅膀鸟类的飞行成本通常更接近实验室预测值，且会因风况不同而出现较大波动。

国际计量局在法国凡尔赛召开会议，多数成员国投票支持一项《关于确保“协调世界时”（UTC）连续性所需的技术措施》的决议，这意味着使用了半个多世纪的闰秒将被取消。根据决议，取消闰秒的举措将于 2027 年 5 月 20 日（世界计量日）生效。此前，在国际计量局 2022 年举行的会议上，成员国投票通过一项决议，呼吁制定不晚于 2035 年逐步取消闰秒的计划。决议还建议用闰时取代闰秒，即等到“原子时”（TAI）和“世界时”（UT）的时刻相差 3600 秒（即1小时）时，再进行调整。闰秒因为需要频繁更改而会导致数字系统发生混乱，闰时则需要 3000 多年才会修改一次。

美国银行（Bank of America）CEO Brian Moynihan 透露该公司雇佣了千名合同工帮助修复 Anthropic Mythos 等 AI 工具发现的软件漏洞。他表示：“我们在第二季度聘请了 1000 名合同工协助进行修复工作。除了原本负责该领域的 3000 名员外，这些合同工至今仍留在公司内，继续提供协助。”他向分析师透露，美国银行第三季度在软件修复以及与反洗钱合规相关的监管问题上投入了数亿美元以上的资金。这是美国大型企业首次披露其为保护软件系统免受 Mythos 影响所投入的人力和资金规模。

加拿大多伦多大学公民实验室发表研究报告，以色列公司 Discover-Tech 通过数字广告在半岛电视台的内部设备上安装了间谍软件。这是首次利用数字广告实施的零点击间谍软件攻击。Discover-Tech 运营着名为 Controlium 的平台，它表面上是一家合法的广告公司，但实际上是以色列间谍软件供应商的合作伙伴。Controlium 能在广告加载的瞬间，对设备进行去匿名化处理、提取设备指纹，自动植入恶意软件。对 Discover-Tech 四十亿条竞价记录的分析显示，在四个月时间里，该公司的广告出现在数十个国家、近 20 万个不同的网站及 Android/iOS 应用中。公民实验室称，名为 DARK CARTOGRAPHER 的组织正利用 Discover-Tech 的基础设施，通过挖掘多家移动运营商系统中的漏洞，将设备的广告标识符与电话号码进行关联，对互联网用户实施大规模的去匿名化。Google 等已采取行动，切断了 Discover-Tech 接入数字广告生态系统的渠道。

Canonical 释出了代号为 Stonking Stingray 的 Ubuntu 26.10，这是一个短期支持版本，仅支持九个月。Ubuntu 26.10 的主要变化包括：用 Rust 语言开发的版本替代了 C 语言开发的系统组件，后量子密码学软件包签名，扩展对基于 TPM 的全盘加密支持，企业级身份验证与 Microsoft Entra ID 的更深度集成，Linux kernel 7.3，新 RISC-V 桌面镜像，GNOME 51 桌面环境，可选启用的本地处理语音转文字方案 Myna，等等。

代表美国科技巨头亚马逊、Google 和 Meta 的行业组织 CCIA Europe 委托进行的一项研究声称，欧洲如果推行数据主权法案 Cloud and AI Development Act（CADA） 会在十年内带来 1150 亿欧元的成本，是欧盟委员会预估的最高值的六倍多。其中 800 亿欧元的成本是假设数字主权法扩大至私营部门关键服务四分之一的云工作负载。报告特别针对了 CADA 中的第 31 条，称其可能允许欧盟委员会将数字主权要求扩大到私营部门的关键服务领域。CCIA 认为强迫企业放弃现有的云基础设施，可能会引发昂贵的软件重开发，其中代码重写和架构重构占预计成本的 69%。

Google 现在会自动验证用户的手机号码，将其关联至用户 Android 设备上登录的所有 Google 账户，全程不会弹出明显确认提示。该功能在 Google 设置中被称为电话号码验证(Phone number verification)，会将用户当前使用的手机号码关联至同一台设备上登录的所有 Google 账户——无论是辅助账户、工作账户，还是其它临时账户。用户通过社交媒体报告收到了 [email protected] 的确认邮件，通知其手机号码已通过验证。调查显示，Google 支持页面于 2025 年 8 月至 10 月间进行了更新，默认启用电话号码的验证功能。用户可通过：设置——Google 或 Google 服务与偏好设置——所有服务——隐私与安全——电话号码验证，关闭该自动电话验证功能。

多项调查数据显示，美国人的社交活动过去三十年大幅减少。American Time Use Survey 的调查数据显示，相比二十年前，美国人独处的时间更多，与朋友相处的时间更少。American Perspectives Survey 的调查发现，有五个或更多密友的美国人比例下降了 25 个百分点，从 63% 降至 38%。Data for Progress 的调查发现，从 1995 年到 2026 年，美国人参与各项关键社交活动的比例出现了显著下降，降幅均达到两位数：去上课或参加研讨会（-22）、在家招待客人（-21）、去教堂或其它宗教场所（-20）、外出野餐（-19）、参加社团聚会（-19）、打牌（-19）、去餐馆吃早餐（-13）以及从事志愿服务（-12）。与此同时使用互联网（+90）、在家玩游戏（+22）、参加健身课程（+14）、音乐会（+13），因个人原因乘飞机出行（+5）。